PERSPECTIVE

Primate color vision: A comparative perspective

GERALD H. JACOBSNeuroscience Research Institute and Department of Psychology, University of California, Santa Barbara, California

(RECEIVED June 18, 2008; ACCEPTED August 28, 2008)

Abstract

Thirty years ago virtually everything known about primate color vision derived from psychophysical studies of normaland color-defective humans and from physiological investigations of the visual system of the macaque monkey, themost popular of human surrogates for this purpose. The years since have witnessed much progress toward the goal ofunderstanding this remarkable feature of primate vision. Among many advances, investigations focused on naturallyoccurring variations in color vision in a wide range of nonhuman primate species have proven to be particularlyvaluable. Results from such studies have been central to our expanding understanding of the interrelationships betweenopsin genes, cone photopigments, neural organization, and color vision. This work is also yielding valuable insightsinto the evolution of color vision.

Keywords: Cone photopigments, Opsin genes, Evolution, Retinal organization, Primate color vision

Introduction

In his classical review of vertebrate eyes, Gordon Walls drewa sharp distinction between the apparently limited color visioncapacities of most mammals and those enjoyed by primates byremarking, ‘‘. . . we come at last to the primate order. Here . . .there has never been any doubt of the occurrence of color vision inall its glory.’’ (Walls, 1942). Having said that, however, he thenwent on to list evidence to suggest that even among the variousprimate lineages, color vision is probably not an invariantcapacity. The species variations hinted in Walls’ early reviewturned out to be more profound than he imagined, and in recentyears studies of these variations have added important detail to thegrowing catalogue of perceptual capacities in various primates.Equally important, this work has also proven valuable in advanc-ing our general understanding of primate color vision in at leastthree ways: (1) by providing deep insights into the early stagebiological mechanisms that support color vision, (2) by fosteringa data-based scenario for how color vision may have evolved, and(3) by setting the stage for a greatly renewed interest in theecology of primate color vision. Here, I review and comment onthe first two of these topics, the third having recently beenaddressed in detail in several publications (Regan et al., 2001;Dominy, 2004; Osorio et al., 2004; Vorobyev, 2004; Jacobs, 2007).

Vertebrate photopigments and their evolution

We have seen that parts many times repeated are eminentlyliable to vary in number and structure; consequently it is

quite probable that natural selection, during the long-continued course of modification, should have seized ona certain number of primordially similar elements, manytimes repeated, and have adapted them to diverse purposes.

Charles Darwin (1859)

Some 45 years ago, application of the newly devised techniqueof microspectrophotometry yielded the first direct measurements ofthe absorption properties of the three types of cone photopigmentof the primate retina (Brown & Wald, 1963; Marks et al., 1964).For all but the hardiest iconoclasts those measurements finallybrought to close long-held arguments about the identity of the first-stage mechanisms underlying primate trichromacy. Through theexploitation of direct measurements of the absorption spectra ofcone photopigments, and from inferences about the absorptionproperties of these pigments derived from the structure of opsingenes, recent years have witnessed an explosion of informationabout the spectra of cone photopigments in a wide range ofvertebrate species. In turn, that information allows the constructionof scenarios for the evolution of vertebrate photopigments.

The consensus view is that all vertebrate visual pigments areproducts of five families of opsin genes (for extensive reviewswith specification of genes and photopigments for a large numberof vertebrate species, see the following: Yokoyama, 2000; Hart &Hunt, 2007; Bowmaker, 2008). One family (termed Rh1) specifiesrod opsins; the expression products of the others (SWS1, SWS2,Rh2, and LWS) are cone opsins. These groups are believed to havearisen through a series of gene duplications. The dating of theseduplication events remains uncertain, but it seems likely that (a)the four cone opsin gene families have an ancient origin, probablyhaving emerged as long as 540 million years ago (mya), and(b) rod photopigments appeared only following the divergence of

Address correspondence and reprint requests to: Gerald H. Jacobs,Neuroscience Research Institute, University of California, Santa Barbara,CA 93106. E-mail: jacobs@psych.ucsb.edu

619

Visual Neuroscience (2008), 25, 619–633. Printed in the USA.Copyright � 2008 Cambridge University Press 0952-5238/08 $25.00doi:10.1017/S0952523808080760

the several cone opsin gene families (Collin & Trezise, 2004).Variations in gene sequence within each of these families yieldsphotopigments that are selectively tuned to absorb maximally indifferent portions of the spectrum. Fig. 1 shows the range of peaksensitivities (�max) for photopigments derived from each of the fivegene families. In some vertebrates, the absorption properties of thephotoreceptors are further modified by variations in the nature ofthe chromophore and/or by intraocular filtering. The effects of thelatter are particularly dramatic in those photoreceptors that in-corporate colored oil droplets; in such cases, as found for instancein the eyes of some birds, reptiles, fishes, and amphibians, theeffective peak sensitivity of the receptor may be shifted as much as60 nm longer than the �max of the photopigment.

Of the major vertebrate groups, pigments derived from all fourof the cone opsin gene families are found in many birds, fishes,and reptiles. Rh2 opsin genes appear to be absent from contem-porary amphibians, while eutherian mammals have neither Rh2opsin genes nor SWS2 opsin genes (Bowmaker, 2008). The timingand circumstances of the loss of these two are not known,although it is usually assumed to have occurred early in mamma-lian evolution, likely coincident with the nocturnal phase ofmammalian history. It has recently been discovered that speciesfrom Marsupalia and Monotremata may constitute exceptions tothis mammalian principle in that three functional classes of conehave been detected in some Australian marsupials, one an SWS1gene, one an LWS gene, and one that may be an Rh2 gene (Arreseet al., 2002, 2006), while the monotreme platypus appears to havetwo functional cone opsin genes (drawn, respectively, from theLWS and SWS2 gene families) plus an additional, but non-functional, SWS1 opsin gene (Davies et al., 2007).

For primates, the important functional consequence of thisaccount of the evolution of vertebrate opsin genes is that the pho-topigments available to support color vision are limited to repre-sentatives drawn from only two of the cone opsin gene families,LWS and SWS1 (Fig. 1).

Evolution of primate cone pigments

As for vertebrates in general, the picture of the evolution of conephotopigments among the primates has been derived from study ofcontemporary species, from inferences gleaned from examinationof fossil material, and from molecular comparisons of cone opsingenes.

Primate phylogeny

Fig. 2 is a simplified primate phylogeny that provides a convenientframework for the subsequent discussion of the evolution of conephotopigments. It represents results from a molecular comparisonof sequences derived from autosomal genes obtained from 13 spe-cies of extant primates (Steiper & Young, 2006). The number ofliving primates (~350 species) vastly exceeds this sample size, butthe animals examined can be considered as representative forpresent purposes. The timings of the divergence events in phylog-enies such as those of Fig. 2 are matters of continuing debate,subject to reinterpretation as new evidences emerge. Generally,analyses of the fossil record lead to more recent divergenceestimates than do phylogenies derived from molecular com-parisons. For example, the earliest available fossil material foreuprimates (‘‘primates of modern aspect’’) dates to ~55 mya,whereas molecular-based estimates, such as those of Fig. 2, oftenplace their origin much earlier—in the Cretaceous, perhaps ~80 mya.A recent statistical analysis designed to account for gaps in thefossil record provides support for the earlier date for the origin ofprimates (Martin et al., 2007), as does a species-level reconstruc-tion of a phylogeny for nearly all the extant species of mammals(Bininda-Emonds et al., 2007).

The earliest primates: Nocturnal or diurnal?

Although the exact timing of the events characterizing primateevolution is not critical for the current understanding of howprimate color vision may have evolved, it is relevant to ask aboutthe photic habits of the earliest representatives of our order. Basedprincipally on an analysis of their sizes, the traditional interpre-tation is that for much of their early history, mammals were smalland nocturnal (Kielan-Jaworowska et al., 2004). Starting fromthat understanding and buttressed by a series of large-scalecomparisons of a variety of physical features of fossil and con-temporary primate material, it has been widely assumed thatancestral primates were themselves nocturnal (Martin, 1990;Heesy & Ross, 2001; Martin & Ross, 2005). In support of thisidea, a recent examination of craniodental remains from a basalAfrican anthropoid dated ~37 mya indicates that it had orbitalfeatures consistent with a nocturnal lifestyle (Seiffert et al., 2005)If early primates were nocturnal, as these observations wouldsuggest, it would seem logical that the cone pigment comple-ments of these earliest primates were those now typical of mostmammals, all of which had a nocturnal past, that is, two types ofcone photopigment, each drawn, respectively, from the SWS1 andLWS opsin gene families (Fig. 1). The spectral positioning ofSWS1-based pigment in early primates is uncertain, but from acomparative analysis of LWS opsin genes from a number of mam-mals, it has been suggested that these animals may have had anLWS pigment with �max of ~553 nm (Yokoyama & Radlwimmer,1999).

There are challenges to the view that the earliest primates werenocturnal. For example, examination of a number of features of askull from a euprimate dated 55 mya has been taken to suggest thatthis animal was small and diurnal (Ni et al., 2003). And a compar-ative examination of cone opsin genes from a variety of strepsir-rhine species has also been interpreted as consistent with the ideathat the ancestral primates were not nocturnal (Tan et al., 2005).Both of these claims have been refuted (Ross & Martin, 2007) andfor now the weight of evidence seems firmly on the side of thosewho believe that our earliest primate ancestors were nocturnal.

Fig. 1. Five opsin gene families specify all the photoreceptor opsins found

in vertebrate retinas. One of these (Rh1) is associated with photopigments

found in rods; the other four link to cone pigments. The extents of the

horizontal lines indicate the total range of �max values for retinal 1–based

pigments specified by each of the five gene families.

620 Jacobs

Evolution and nature of catarrhine cone pigments

Unlike most mammals, representative catarrhine species (a groupmade up of Old World monkeys, apes, and humans; Fig. 2) havetwo X-chromosome cone opsin genes. These genes are locatedadjacent to one another on the X-chromosome and are highlyhomologous, indicating that they probably emerged as a result ofa recent gene duplication (Nathans et al., 1986). The fact that thecatarrhine gene arrangement is distinctive from that characteristicof the platyrrhine (New World) monkeys (below), and that allcatarrhines seem to share the same arrangement, implies that thisduplication occurred following the platyrrhine–catarrhine diver-gence, but prior to the diversification of the catarrhines, at perhaps~35 mya (Fig. 2). Although there are numerous catarrhine speciesthat have not been studied, those that have (in total, animals drawnfrom at least 10 genera) all seem to have separate classes of conecontaining M (middle-wavelength sensitive) and L (long-wavelengthsensitive) pigments with respective �max values of ~530 and 560 nm(Bowmaker et al., 1991; Jacobs, 1996; Jacobs & Deegan, 1999).

The retinas of all catarrhines also have a small population ofphotoreceptors (some 5%–10% of all cones—Calkins, 2001)containing S (short-wavelength sensitive) pigment. The opsingene specifying this pigment is localized to chromosome 7. Thecurrent view is that the ancestral mammalian SWS1 pigment wasmaximally sensitive in the ultraviolet (UV). In primates, as inmany other mammalian lineages, a combination of substitutions in

the opsin sequence shifted the spectral positioning of the pigmentsuch that it absorbs maximally in the short-wavelength portion ofthe visible spectrum (Hunt et al., 2004). When that event may haveoccurred is not known. There are relatively few direct measure-ments of the absorption properties of primate S pigments, butthere is reason to believe that, at least, there may be a modestdifference between the human S pigment and that for variouscatarrhine monkeys with the human pigment being positionedrelatively shorter (respective �max values of ~430 and 415 nm)(Bowmaker, 1990). Absorption spectra for the nominal three typesof cone pigments found in catarrhine primates are sketched inFig. 3A.

Catarrhines effectively share their opsin gene and cone photo-pigment complements, but there are several known speciesidiosyncrasies, and there are probably more yet to be discovered.The most prominent of these is the presence of gene/pigmentpolymorphisms in humans that are effectively absent in othercatarrhines. The human polymorphisms encompass both the large-scale changes collectively linked to color-defective vision anda number of alterations that seemingly have little impact onnormal vision (Neitz & Neitz, 2003). The former include theabsence of one or the other of the normal M or L pigment(resulting in dichromatic color vision) and significant shifts in thespectral positioning of one of the M or L pigment (yieldinganomalous trichromacies). Together, these polymorphisms affect~8% of all Caucasian males. Though they have been less well

Fig. 2. Phylogeny of the major primate groups. The divergence times were estimated from a Bayesian analysis of genomic data obtained

from 13 species of extant primates. So derived, the last common ancestor of modern primates dates to ~77 mya, while the two major

groups of anthropoids, platyrrhines (New World monkeys) and catarrhines (Old World monkeys, apes, and humans), diverged ~43 mya.

Modified from Steiper and Young (2006).

Primate color vision: A comparative perspective 621

scrutinized, analogous changes are, at best, infrequent in othercatarrhines. That fact is made clear in Table 1, which documentsthat the incidence of dichromacy in a large population of humanmales is strikingly higher than that in a collection of malemacaque monkeys. Although a number of suggestions have beenoffered as to the origin of this dramatic difference between theselineages, including a well-known early proposal that it may haveresulted from a relaxation of selection against defective colorvision in modern humans (Post, 1962), there is as yet no agreedexplanation for this difference (Jacobs & Williams, 2001).

Evolution and nature of platyrrhine cone pigments

A case, of sorts, could be made for considering thattrichromacy has evolved independently in the catarrhinesand platyrrhines.

Gordon L. Walls (1942)

As for the catarrhines, many platyrrhine lineages (Fig. 2) havesucceeded in moving beyond the mammalian norm of a singletype of M/L cone pigment. In the case of these monkeys, however,the additional pigments generally derive not from a gene dupli-cation event, as they do in the catarrhines primates, but rather fromthe emergence of opsin gene polymorphisms (for reviews, seeJacobs, 1998, 2007). These polymorphisms are at an X-chromosomeopsin gene locus, thus allowing individual animals to have varyingopsin gene and cone photopigment complements. The typicalarrangement features three opsin gene alleles, each specifying anM/L pigment with a unique spectral absorption property. In-dividual monkeys have any one of the three and, along with theproduct of an S-cone gene (as for catarrhines, mapped to chromo-some 7), their retinas contain a total of two types of cone pigment,or they have any pair of the M/L pigments which along with anS pigment gives them a total of three types of cone pigment. Sincethere is only a single opsin gene on any X-chromosome in theseplatyrrhines, to gain a total of three cone pigments requires thatdifferent genes be present on two X-chromosomes, that is, theoccurrence of three photopigments is restricted to females whoare heterozygous at the opsin gene site. The net result of thisarrangement is that six different photopigment phenotypes arerepresented in these species.

Although most platyrrhine monkeys have opsin gene poly-morphisms of the sort just described, they are not all the same.One source of variation is that different species have unique setsof M/L opsin genes, thus providing variations in the spectralabsorption properties of the M/L pigments. Two patterns pre-dominate. The three types of M/L pigments for species from thefamily Cebidae (Cebus and squirrel monkeys) feature �max valuesof ~535, 550, and 562 nm. By contrast, the three types of M/Lpigments found in species of the family Callitrichidae (marmosets,tamarins, etc.) have pigments with �max values of ~543, 556, and562 nm, respectively. Although they have been less intensivelyinvestigated, the cone pigments found in members of several otherplatyrrhine families tend to follow the latter pattern (a list ofthe platyrrhine pigments so far measured with accompanying dis-cussion of the measurement details appears in Jacobs, 2007).Evidence also suggests that there are some variations in thenumber of polymorphic genes and pigments. For example, threegenera (spider monkeys—Ateles, pygmy marmosets—Cebuella, andGoeldi monkeys—Callimico) appear to have two not three M/Lpigment variants (Jacobs, 2007). This apparent reduction could beartifactual, reflective of nothing more than sampling problems, butif it is real this arrangement significantly limits the number andvariety of pigment combinations present in these monkey pop-ulations. The other deviation of this type occurs in monkeys fromthe genus Callicebus (titi monkeys). These primates appear tohave an embarrassment of pigment riches featuring a total of fivealternative versions of the M/L pigments (Jacobs & Deegan,2005), an arrangement that should also greatly affect the totalnumber of available pigment combinations in the population.

As is true for the catarrhines, platyrrhine S-cone pigments havenot yet been subject to extensive study. The available mea-surements suggest that platyrrhine S cone pigments have �max of

Fig. 3. Spectral sensitivity functions for primate cone photopigments.

(A) Three types of cone pigment characteristic of the catarrhine primates.

(B) Cone pigments of platyrrhine monkeys from the family Cebidae. (C) Cone

pigments of platyrrhine monkeys from the family Callitrichidae. Pigments

sketched as solid lines are those characteristically found in all individuals;

the polymorphic M/L pigments appear as dashed lines. The �max values for

all these cone pigments are given in the text.

Table 1. Assessments of the incidence of dichromatic color visionin two primate populations

Population Sample size (n) Prevalence (%)

Human males 20,836 2.56Macaque males 1629 0.002

The human results were drawn from several large surveys conducted onCaucasian males that are cited in Fletcher and Voke (1985). Results for themacaque monkeys were derived from Jacobs and Williams (2001) andOnishi et al. (1999).

622 Jacobs

~430 nm with, possibly, some modest variation from this figure indifferent species. The absorption spectra for the two most prev-alent patterns of cone pigments seen in the polymorphic platyr-rhine monkeys are shown in Fig. 3B and 3C.

Two significant deviations from the polymorphic theme havebeen detected in platyrrhine monkeys. The first is in Aotus, the owlmonkey. This monkey is unique in being the only anthropoidconsidered to be nocturnal. Aotus lacks the M/L opsin polymor-phism, having a single type of M/L pigment (�max of ~543 nm).Even more intriguing, owl monkeys express no viable S-conepigment with the result that their retina contains only a single conetype. The absence of S cones in Aotus results from the presence ofmutational changes in the S-cone opsin gene that obviate proteinexpression (Jacobs et al., 1996b). The fact that this same change isfound in several different species of Aotus implies that the loss ofS cones must have appeared early in the history of this lineage(Levenson et al., 2007). Interestingly, in recent years, a loss ofS cones having a similar genetic basis has also been documented ina number of other mammalian species (reviewed in Peichl, 2005).The mammals comprising this group occupy greatly disparate en-vironmental niches and have varied evolutionary histories, makingit uncertain as what event(s) may have precipitated these gene-driven losses of a cone type. Whatever the reason, the compellingfunctional consequence is that these mammals must lack anyconventional color vision capacity.

A second deviation from the platyrrhine polymorphic patternhas been detected in monkeys from the genus Alouatta (howlermonkeys). Surprisingly, these monkeys resemble the catarrhines inroutinely having two X-chromosome opsin genes, so providingeach individual with retinas containing separate populations of

M and L cones (Jacobs et al., 1996a). The M and L cone pigmentsof howler monkeys have spectral absorption properties verysimilar to those of catarrhine primates, that is, �max values of~530 and 560 nm (Saito et al., 2004).

The sequences for M/L opsin genes are very similar across allprimates, for example, the M/L pigment genes for a catarrhine (thehuman) and two species of platyrrhine monkeys (squirrel andmarmoset monkeys) share a sequence identity of 96% or greater(Neitz et al., 1991; Hunt et al., 1993). This high similarity in theface of variations in the spectral positioning of the M/L pigmentsindicates that only a small number of changes in gene structuremust be responsible for variations in spectral tuning. In fact, itappears that most of the variation in spectral tuning over theprimate M/L pigment range of ~30 nm is due to amino acidsubstitutions at only three sites in the opsin molecule (positionsnumbered 180, 277, and 285 in the diagram of Fig. 4). In eachcase, the replacement of a nonpolar amino acid with hydroxyl-bearing amino acid (or vice versa) yields a discrete shift in thepeak of the pigment with the total shift of the pigment being ef-fectively additive for changes at the three sites. The net effect of thiscontrol for spectral tuning is that the nonhuman primate M/L pig-ments occupy only a small number of spectral locations. Table 2documents this fact for representative genera drawn from all threemajor groups of primate.

The conservatism of spectral tuning of the primate M/Lpigments provides the basis for drawing inferences about theevolution of the platyrrhine cone pigments. Comparisons of thesequences of the M/L opsin genes in the platyrrhines suggest thatthe polymorphisms characteristic of these species are of longstanding, extending well back into platyrrhine history (as much as

Fig. 4. Schematic representation of the primate M/L cone opsin. The amino acids are indicated by small circles. The opsin threads its

way through the disc membrane of the cone outer segment in a series of seven linked helical arrays. The amino acids that are

principally responsible for shifting the spectral absorption properties of the cone pigment are at the positions numbered 180, 277, and

285. Substitutions at those sites shift pigment absorption by approximately the amounts indicated in the inset; for example, substituting

alanine for serine at position 180 shifts the pigment ~6 nm toward the shorter wavelengths. The retinal 1 chromophore is covalently

bound to opsin at residue 312.

Primate color vision: A comparative perspective 623

20 mya), and that their evolution is better explained as havingproceeded along lines more specific to the spectral positioning ofthe pigment rather than to strict species lineages (Boissinot et al.,1998; Surridge & Mundy, 2002). Howler monkeys with their twoX-chromosome opsin gene sites obviously had a different history.As in the catarrhines, that must have included an X-chromosomeopsin gene duplication. When the duplication may have emergedis uncertain, but the fact that it is unique to Alouatta means itoccurred after this line had diverged from its sister clades, that is,more recently than ~12 mya (Schrago, 2007). Comparison of theM/L gene sequences in the howlers with those of catarrhineprimates also strongly implies that the respective gene duplicationswere independent events, having occurred much more recently inthe howler monkeys than in the catarrhines (Kainz et al., 1998).

Evolution and nature of strepsirrhine cone pigments

The suborder Strepsirrhini is composed of seven families ofprimates that are collectively native to Madagascar, SoutheastAsia, and Africa. The eyes of the several dozen species that makeup this diverse group of primates (Fig. 2) share a number offeatures that serve to sharply differentiate them from the eyes ofanthropoids; among these, strepsirrhine retinas lack foveae andfeature lowered cone:rod ratios, and their eyes often containreflective tapeta. In the face of these generally more primitivefeatures, these animals show a surprising range of cone photo-pigment arrangements. Although this story continues to develop,three general patterns have been detected to date. The firstresembles that described for Aotus, that is, featuring a singlefunctional pigment from the LWS gene family along with anS-cone opsin gene that has sufficient mutational changes to renderit ineffective. All the lorises and bush babies appear to share thisarrangement, along with a few species of lemur (Tan & Li, 1999;Kawamura & Kubotera, 2004; Tan et al., 2005). In a single casewhere it has been measured, the M/L cone pigment in such specieshas �max of ~543 nm (Jacobs et al., 1996b). Second, somestrepsirrhines, ring-tailed and brown lemurs being common ex-amples, have both functional S cones and a single type of M/Lcone, an arrangement qualitatively similar to that of the mamma-lian norm (above). The spectral peaks of the pigments in animalsof this type have an M/L pigment with a peak similar to those ofthe lorises and bush babies, that is, ~545 nm, while the S pigmentpeak in one measurement was at ~437 nm (Jacobs & Deegan,1993). The third cone pigment pattern is generically similar to thatseen in the polymorphic platyrrhines. Such animals have twoalternative versions of M/L opsin gene and photopigments, alongwith an intact S-cone opsin gene (Tan & Li, 1999). This yieldsthree different photopigment phenotypes to include heterozygousfemales whose retinas contain three separate cone types. TheM and L cones of such animals have respective �max of ~545 and

558 nm, while S cones have peak values of ~430 nm (Jacobs et al.,2002). Individual species from two strepsirrhine families (Lemur-idae and Indridae) have thus far been found to share this arrange-ment, but it probably also exists in additional species from thesetwo families. These three varying patterns of strepsirrhine conepigment arrangements are sketched in Fig. 5.

Among the strepsirrhines, the highly endangered aye-aye(Daubentonia) undoubtedly takes the prize as the most unusual.Examination of the opsin genes in this profoundly nocturnal

Table 2. Amino acids involved in spectral tuning of M/L pigments in several genera of nonhuman primates

Peak absorption (nm) Genus* Position 180 Position 277 Position 285

562 Cebus, Saimiri, Callithrix, Saguinus, Alouatta, Macaca Serine Tyrosine Threonine556 Callithrix, Saguinus, Aotus, Propithecus Alanine Tyrosine Threonine550 Cebus, Saimiri Alanine Phenylalanine Threonine543 Callithrix, Saguinus, Propithecus Alanine Tyrosine Alanine535 Cebus, Saimiri, Alouatta, Macaca Alanine Phenylalanine Alanine

*Catarrhine: Macaca; platyrrhine: Cebus, Saimiri, Callithrix, Saguinus, Alouatta, Aotus. strepsirrhine: Propithecus.

Fig. 5. Spectral sensitivity functions for cone pigments of strepsirrhine

primates. The three panels reflect the three patterns described in the text.

(A) A single type of M/L cone pigment. (B) An S-cone pigment and a type

of M/L pigment. (C) An S-cone pigment and polymorphic versions

(dashed lines) of two types of M/L cone pigments. Identities of the spe-

cies having these variant patterns and the pigment �max values are given in

the text.

624 Jacobs

primate suggests that the retina should have two functional classesof cone pigment (Perry et al., 2007). Such an arrangement wouldnot be unusual, but a recent brief report suggests that the residuespresent at tuning sites in the aye-aye SWS opsin gene predict thatthe short-wavelength cone pigment may have its absorption peakin the UV, not in the visible wavelengths (Hunt, 2006). If true,this would be a unique arrangement for a primate, and that factalone should make this species an important target for furtherexamination.

Finally, a word about tarsiers. Tarsiers are curious Asianprimates whose taxonomic position has been a source of muchdebate. Although classically lumped together with the othersconsidered in this section (as being, collectively, ‘‘prosimians’’),the modern cladistic consensus (Schmitz et al., 2001) is thattarsiers are now properly considered as members of the suborderHaplorhini, apparently more closely related to the anthropoidsthan to the strepsirrhines (Ross & Martin, 2007). These smallnocturnal primates are primitive in many respects, and althoughthey are not included in the primate phylogeny derived for Fig. 2,they have enjoyed an extended period of evolution independentfrom that of other primate lineages. Recent immunocytochemicallabeling experiments identify separate populations of S and M/Lcones in the retina of one species of tarsier (Hendrickson et al.,2000), and this suggests that they may have an opsin gene andcone pigment complement similar to that described above for thering-tailed and brown lemurs.

If the basal condition for primates included an S and a singleM/L opsin gene coding for two classes of cone pigment, then atleast two types of change must have occurred in the evolutionof strepsirrhine lineages: the loss of a population of functionalS cones in many species and the acquisition of M/L polymorphismin some others. The timing of the former is uncertain; however, themutational changes that render the S-cone opsin gene nonfunc-tional in lorises and bush babies are shared, implying that at leastin this case, the loss occurred long ago in an ancestor that wascommon to the two lineages (Kawamura & Kubotera, 2004).When M/L opsin gene polymorphisms arose in the strepsirrhinesis even less clear, but establishing that fact could be important fordetermining the course of the evolution of primate trichromacy asdiscussed in the next section.

History of primate opsin gene polymorphism

As noted, sequence comparisons suggest that X-chromosomeopsin gene polymorphisms appeared early in the evolution ofplatyrrhine monkeys. The recent discovery that similar poly-morphisms are found among the strepsirrhines raises the questionof whether such polymorphisms arose independently in theplatyrrhines only after they had diverged from the catarrhines(Fig. 2) or whether polymorphism is a much more ancient featureof primate visual system evolution. The following observationsmay be relevant to deciding this issue.

(1) On the basis of sequence comparisons of strepsirrhine opsingenes, Tan et al. (2005) suggested that the common ancestorof all primates was polymorphic for the M/L alleles. If thatwere the case, then opsin gene polymorphism has probablyexisted throughout primate history, and so its appearance inplatyrrhines and some strepsirrhines is simply a contempo-rary manifestation of a long-standing feature. Since it is noteasy to see how M/L pigment polymorphism would promotefitness in nocturnal animals, this claim runs counter to the

standard view (above) that ancestral primates were nocturnal.But if, as Tan et al. suggest, polymorphism in fact did ariseearly in primate evolution, then it was subsequently lost in(a) many lineages that are currently nocturnal, (b) others thatare not currently nocturnal but do express only a single M/Lcone pigment, and (c) those species that have acquired a secondX-chromosome opsin gene.

(2) The same spectral tuning sites are used for all nonhumanprimate M/L opsin genes (Table 2), and this coincidence issometimes taken to suggest that they all may have hada common origin. Against that possibility, it may be that onlya handful of amino acid variations can be used to generateM/L spectral absorption differences, and thus their commonappearance in different lineages merely reflects not commonancestry but rather convergent evolution.

(3) There is less nucleotide divergence among the allelic formsof the platyrrhine M/L genes then there is between thecatarrhine M and L genes (Hunt et al., 2005). That factwould suggest that the platyrrhine polymorphism is morerecent than the catarrhine gene duplication and thus thatpolymorphism did not exist among the early catarrhines.However, sequence homogenization leading to interallelicrecombination might also produce this same effect, and thusthis difference may not be compelling one way or another.

Taking all these facts together, it seems clear that at presentthere are sufficient contradictory indicators that make it impossi-ble to decide whether photopigment polymorphism arose early inprimate evolution, and then was lost at various junctures, or whetherpolymorphism emerged only at an early stage in platyrrhine di-versification and also, independently, in the more distal branchesof the strepsirrhine radiation.

M/L photopigment expression

To provide receptor signals optimal for supporting color vision,each photopigment type needs to be individually expressed incone receptors. In primates having only a single type of pigmentgene on the X-chromosome (e.g., all platyrrhine male monkeys),no problem arises. However, those primates that harbor more thanone type of M/L pigment gene require additional specification.One key to assuring individual expression of different pigmenttypes is the dosage compensation mechanism of X-chromosomeinactivation, the general control device that has evolved to equal-ize protein expression in male and female cells in which, duringearly embryological development, one of the two X-chromosomesis randomly inactivated in female cells, and this inactivation is main-tained through all the progenies of each cell. For the heterozygousfemales of those species that are polymorphic, X-chromosomeinactivation neatly solves the problem of selective pigment ex-pression in that one of the two M/L pigments is expressed inroughly half of her cones, while others express the other pigmenttype (Jacobs & Williams, 2006).

With two different genes on the X-chromosome, catarrhineprimates require an additional control to assure the individualexpression of cone pigment. That mechanism apparently involvesinteractions with a region of DNA located upstream to the opsinarray, the so-called locus control region (LCR). A series ofexperiments have established that the LCR pairs with the promotersequences of either the M or the L gene and, in so doing, de-termines which gene gets expressed in any given cone cell (Wanget al., 1992; Smallwood et al., 2002). That pairing (illustrated in

Primate color vision: A comparative perspective 625

Fig. 6A) is believed to reflect a random process and is argued to bethe basis from which the position of the M and L cones in theretinal mosaic is, at least across local regions, also effectivelyrandom. Beyond the catarrhines, only one other primate, thehowler monkey, is known to have two opsin genes on the X-chromosome. The means for expression of M/L cone pigments inthese monkeys is less clear, because in these animals rather thanthere being a single upstream LCR, an LCR is associated witheach of the pigment genes (Dulai et al., 1999) (Fig. 6B). Just howthis arrangement allows selective receptor expression of the M andL pigments is not known. One possibility, apparently so farunevaluated, is that the downstream LCR has, over time, accu-mulated changes that render it currently nonfunctional.

X-chromosome opsin gene duplication

Gene duplication has long been recognized as a process central toproviding the basis for the evolution of new gene function (Zhang,2003). X-chromosome opsin gene duplication must have occurredat least twice during primate evolution—at the base of thecatarrhine radiation and in an ancestor to present-day howlermonkeys. The tandem arrangement of the catarrhine L and M coneopsin genes suggests that they emerged as a result of unequalcrossing-over. Logically, this process could have placed twoidentical cone opsin genes on an X-chromosome, and these twothen would have subsequently diverged in structure so as to codefor separate M and L opsins. Alternatively, crossing-over mighthave occurred on the background of a preexisting polymorphicarrangement such that separate M and L genes got co-located onthe X-chromosome in a single step. The former process (duplica-tion, then divergence) would be the more typical interpretation(Zhang, 2003), but if polymorphism arose early in primateevolution, as discussed above, then divergence may well havepreceded the duplication step. It seems almost certain that the

latter did occur in howler monkey evolution since, as noted above,this monkey features duplicate LCRs in the X-chromosome opsingene array, and such an arrangement suggests that the processof unequal crossing-over must have involved preexisting allelicforms of the M and L opsin genes.

Evolution of primate color vision

It has long been appreciated that the presence of multiple types ofcone pigment is necessary to support a color vision capacity, but itis not sufficient. In addition, the visual system must be organizedin a manner that allows for the neural comparison of signals orig-inating in cones containing different pigment types. This sectionconsiders some issues linking primate cone pigments to primatecolor vision.

Inferring the dimensionality of color vision

The first clear statement about the dimensionality of color visionis generally said to be contained in Thomas Young’s famous 1802speculation that the number of fundamental mechanisms un-derlying human color vision is limited to three (Mollon, 2003).Color mixing experiments conducted later in the 19th centuryprovided an extensive empirical basis for this claim. Thisfundamental characteristic of human color vision has since thattime been described as trichromacy. Directly linking thesefundamental mechanisms to the presence of three types of conephotopigments is a more recent accomplishment, as was notedabove. Early studies of human color defectives by Konig andothers revealed individuals whose color matches required two notthree fundamentals, and these people were termed dichromats(Smith & Pokorny, 2003). Although dimensionality, be it tri-chromatic or dichromatic, is ultimately defined by color matchingmeasurements, a host of studies of color vision have identifiedother behavioral indices that covary with dimensionality, forexample, results from various types of spectral discriminations,and these are often taken as secondary means to diagnosedimensionality.

Directly measuring color vision in nonhuman subjects isa tedious business usually involving extended periods of trainingand testing. Because of the compulsive linkage between thenumber of cone pigment types and the dimensionality of colorvision in humans, and because it is relatively more efficient toassess the number of cone types either by direct measurement orby inference from studies of cone opsin genes, it is attractive toautomatically assume that the same linkages hold in other species.Enough so that in contemporary reports it has become routine torefer, for example, to evidence for the presence of two cone typesas automatically implying dichromatic color vision. At leastamong the primates, such inferences seem mostly well justified,but there is still room for caution.

For a representative nonhuman catarrhine, the macaque mon-key, and for a few species of the polymorphic platyrrhines (e.g.,squirrel monkeys, marmosets), there are direct behavioral meas-urements to compellingly link the presence of two or three typesof cone pigment to dichromatic and trichromatic color visioncapacities. And the owl monkey, the single anthropoid speciesknown to have only one type of cone pigment, has been shown tobe monochromatic (Jacobs et al., 1993). For the strepsirrhines, thestory is a little less clear-cut. There have been few directbehavioral studies of color vision in these primates and none thatwould seem to provide compelling tests of dimensionality that can

Fig. 6. Illustration of mechanisms proposed to control expression of

X-chromosome opsin genes. (A) In catarrhine primates, the L and M opsin

genes are located in tandem array on the X-chromosome. Experiments

suggest that that an upstream LCR pairs randomly (arrows) with the

promoters for either the L-cone opsin gene (top) or the M-cone opsin gene

(bottom) to assure individual expression of the two pigment types. (B) The

platyrrhine howler monkey also has L and M opsin genes arranged in

a tandem array on the X-chromosome. In this case, there are, as illustrated,

two LCRs; it is not known how these two interact with the opsin gene

promoters to assure individual expression of pigment into the two cone

types (see text discussion).

626 Jacobs

be directly correlated with known photopigment complements.Although it seems likely that the dimensionality of their colorvision aligns with the photopigment story outlined above, the cleardifferences in the organization of anthropoid and strepsirrhineretinas suggest that it would be very useful to have this correlationtested directly, ideally in members of one of the species that havea photopigment polymorphism.

A puzzling primate with regard to linking cone complement tocolor vision is the tarsier. These small animals, variously de-scribed as being nocturnal or crepuscular, have enormous eyeswith a heavily rod-dominated retina that nevertheless displaysa clear fovea. Immunocytochemical labeling experiments identifytwo classes of cone in the tarsier retina—one UV/S and the otherM/L (Hendrickson et al., 2000). There is a significant centroper-ipheral cone density gradient with an almost reciprocal distribu-tion of the two cone types such that there are very few UV/S conesin the central retina, which then become denser near the retinalperiphery, just where the M/L cones are relatively sparselydistributed (Hendrickson et al., 2000). Since extraction of a colorsignal is classically conceived as depending on a local comparisonof inputs from the two cone types, it is not at all clear that thisunusual arrangement of the two cone types will support much ofa color vision capacity, and thus it may be a bit of a stretch toassume that the tarsier is dichromatic. Although the tarsier isundoubtedly unusual, that case may serve as a clear warning that itis not always a given that one can straightforwardly infer colorvision from indicators of the cone pigment complement alone.That caution is particularly apt for cases involving ancestralspecies for whom development of an opsin gene phylogeny maymake it possible to infer the number of cone pigments while at thesame time offering no insight into the number and retinaldistribution of the cones, nor indeed of any of the other detailsof retinal construction.

Rod signals and primate color vision

To this point, the discussion has focused exclusively on variousissues related to the operation of cone photoreceptors. That is inaccord with the traditional linkages made between cone signalsand color vision. But there is a considerable span of photic lightlevels over which rods and cones are jointly operative—in humanobservers, for instance, this domain (mesopic vision) encompassesa range of as much as four log units of light intensity (Wyszecki &Stiles, 1982). Rod and cone signals share output pathways fromthe retina, and thus it is probably not surprising that under mesopictest conditions rods have been demonstrated to influence colorvision in various ways. These influences are complex, perhapsreflecting the multiple sites available for interaction of rod andcone signals in the retina, and they are not easy to economicallysummarize, at least partly because they vary depending onnumerous viewing parameters such as stimulus size, timing, andretinal location (Buck, 2003). Among the most studied of the rodinfluences on color vision are the biases they can exert onperceived hue; for instance, in human observers tested at mesopiclight levels, rod signals preferentially enhance the percept of bluerelative to yellow (Buck et al., 1998). A recently-devised psycho-physical technique predicated on the ability to independentlycontrol rod and cone stimulation offers the promise of being betterable to link appearance measurements such as these to underlyingphysiology, thus ultimately helping to specify how rod and conesignals interact (e.g. Cao et al., 2005). Finally, it has been apparentfor some time that rod signals can also significantly impact the

dimensionality of color vision; for example, with relatively largetest fields, human dichromats behave trichromatically, that is, theymake unique color matches that require the use of three rather thanthe two primaries that suffice when they are tested at photopiclight levels (Smith & Pokorny, 1977).

Although there is thus the clear possibility for rod influenceson color vision, there is virtually no indication of how thesemay manifest themselves in real-world viewing. The opportu-nity is certainly there given that almost all nonhuman primatesare behaviorally active under some conditions of illuminationthat will be mesopic for their eyes—for instance, during thetwilight transitions between day and night and for cases wheresignificant amounts of ambient light are naturally occluded asduring periods of heavy overcast or at deeply shaded locationsin the subcanopy regions of rain forests. Rod influences on colorvision may be even greater for those primates whose visualsystems feature certain organizational features, for instance,relatively greater rod representation, the absence of a fovea,reduction in the number of cone types. With regard to this lat-ter item, it is noteworthy that electrophysiological recordingstudies conducted on both dichromatic and monochromatic plat-yrrhine monkeys reveal the presence of continued rod contri-butions to central visual system neurons at very much higherlight levels than those typical found in the visual systems of tri-chromatic primates (Yeh et al., 1995; Silveira et al., 2004). Whetherthese potential rod influences on color vision are employedadaptively in the conduct of normal primate behavior, and thusmay render the mechanisms underlying rod operation subject toselection, or whether they are simply unavoidable by-products ofthe organization of duplex visual systems is unclear. In sum,although at present not much can yet be said about rod influenceson primate color vision in a comparative context, there is a possibilitythat this topic may comprise an important story still to be written.That point may be particularly apt for the issues covered next.

Nocturnality and primate color vision

A number of strepsirrhine species, as well those from the an-thropoid genus Aotus, have usually been described in the literatureas nocturnal. As noted above, two different cone pigment patternshave been documented in such primates: (1) an SWS1 and anLWS pigment, both seemingly fully functional, and (2) an LWSpigment but no functional SWS1 pigment. The latter arrangementwill not permit color vision, but the former might. Although somespecies (hawk moths and nocturnal geckos) have recently beenshown to be capable of making color discriminations at very lowlight levels, this capacity rests on structural features of their eyesthat allow for a sacrifice of temporal and spatial resolution inexchange for a range of spectral resolution that extends down tovery low light levels (Kelber & Roth, 2006). Such adaptations arenot obviously available in any mammalian eyes, and thus, primatecolor vision is necessarily limited to light levels that are higherthan those present under typical nocturnal conditions. Given thislimitation, why would a nocturnal primate maintain a capacity forcolor vision?

One obvious answer is that the linkage between light level andbehavioral activity in animals is seldom as constrained as textbookdefinitions of nocturnal, diurnal, and crepuscular patterns mightsuggest. In addition to that, it is well recognized that the activitypatterns of various primates are not well accounted by any of thesetraditional divisions but rather are best described as ‘‘cathemeral,’’a term coined to describe activity that may be distributed throughout

Primate color vision: A comparative perspective 627

the 24-h cycle with the particulars of the pattern dependent ona wide variety of factors such as ambient temperature, food avail-ability, predation pressure, etc. (Tattersal, 2006). One of the strep-sirrhines known to be cathemeral is the brown lemur (Eulemurfulvus), an animal that has two functional classes of cone pigment(Jacobs & Deegan, 1993). Clearly, primates such as this willroutinely encounter light levels where color vision could be a visualasset.

There are other primate species classically considered to benocturnal that have two classes of cone pigment and thus at leastthe potential for dichromatic color vision. The opportunities forexploiting this color capacity seem on the face much reduced insuch animals. But they are not absent entirely. Although this willvary from species to species, for humans the luminance of a whitepaper viewed in full moonlight is about double the cone threshold(although perhaps still not high enough to reach the threshold forcolor vision), and there is the suggestion that under such con-ditions, bright saturated colors may be visible (Makous, 2004).Thus, there is at least the marginal possibility that some ‘‘nocturnal’’conditions may support limited color vision. Probably morerelevant, it is well documented that even the most devoutlynocturnal primates will occasionally awaken and become activeduring daylight hours; for example, in response to the presence ofpredators, to respond to weather contingencies, or simply toinitiate searches for sustenance in times of food shortage (Bearderet al., 2006). Such circumstances, even if relatively infrequent,may be sufficient to support the maintenance of a minimal colorvision capacity.

The above examples suggest that there are no facile linkagesto be made between the presence of color vision and photicactivity cycles. That conclusion is made even stronger byobservations on animals that have the second of the two arrange-ments, that is, only a single type of cone pigment and a consequentabsence of color vision. One such example is the owl monkeyAotus azarae. This species, like others of the same genus, lacksa functional SWS1-based pigment (Levenson et al., 2007). Sur-prisingly, although most species of Aotus are nocturnal, A. azaraeis cathemeral, often seen to be active during daylight hours as wellas during the night (Fernandez-Duque, 2003). The daylight lightlevels encountered by this monkey are such that its vision will bemostly based on photon capture by a single cone photopigment.Interestingly, the daytime diet of A. azarae includes, among otherthings, brightly colored tree flowers (Gimenez & Fernandez-Duque, 2003). Observers of such foraging behavior might betempted to conclude that the flower harvest is guided by colorcues, perhaps then proceeding to predict what array of conepigments might best subserve this discrimination. Of course, thismonkey has no capacity to exploit such spectral cues and thusthis case underlines the difficulty of inferring a capacity for colorvision solely from observations of normal behavior. It additionallymakes clear that a complete lack of color vision need not presentan insurmountable barrier to primate success under daylightillumination.

Given the prospect that nocturnal primates are sometimesin environmental circumstances that support the use of colorvision, perhaps the real question is not why some nocturnalprimates have maintained color vision but, rather, why somelineages have seemingly abandoned that possibility through theaccumulation of mutational change in their SWS1 opsin genes?Although the loss of one class of functional cones does not needto have been adaptive, its spread to include all members of thespecies allows the possibility that it may have been. But if so, it

is not obvious how this loss might enhance primate fitness andso, at present, this gene-induced abandonment of a potentialdimension of color vision in select primate lineages remainsvery much a mystery.

Retinal pathways supporting primate color vision

The extraction of color information is initiated in retinal circuitsorganized to contrast photon absorption patterns in cones con-taining different types of pigment. In mammals, two types of suchcircuits have been identified (for reviews, see Martin, 1998;Masland, 2001; Lee, 2004; Wassle, 2004). One incorporatesa dedicated class of bipolar cells that selectively receive inputsfrom S cones. Signals from these bipolar cells are fed to smallbistratified ganglion cells, which also receive inputs from a groupof bipolar cells that contact M/L cones. Signal inputs from thesetwo sources are combined in an antagonistic fashion to form thebasis for the blue/yellow opponent pathway. Although the numberof species in which this pathway has been definitively dem-onstrated thus far is limited (e.g., catarrhine and platyrrhineprimates: Dacey & Lee, 1994; Silveira et al., 1999; some rodents:Haverkamp et al., 2005; Li & DeVries, 2006; and rabbit: MacNeil& Gaul, 2008), it seems reasonable to suppose that this arrange-ment is characteristic of the eutherian retina, implying that it hasbeen maintained throughout the course of most of mammalian evo-lution. Since early mammals are assumed to have been nocturnal,the apparent antiquity of this pathway adds weight to the ideaexplored above that even for animals that are principally noc-turnal, some color capacity provides an adaptive advantage(although, it should be said, there may be reasons beyond supportfor color vision for maintaining two classes of cone). For con-temporary mammals that maintain separate populations of S andM/L cones, these ganglion cells provide the initial neural substratefrom which a single dimension of color vision can emerge. Thusfar, we do not seem to know the fate of this pathway in thoseprimates, such as the Aotus monkey, that have lost their S cones toopsin mutation.

The second retinal circuit for extracting color informationis unique to primates. It originates as M or L cone inputs tomidget bipolar cells, and these in turn synapse on midget ganglioncells (Martin, 1998). Classically, this arrangement was conceivedto provide signal input from a single type of foveal cone (M or L)to a single ganglion cell, which was then combined antagonis-tically with signals that originate from activation of M and Lcones in neighboring regions to form the red/green opponentpathway. There is now evidence that signals from more thanone foveal cone may be collected by an single ganglion cell(McMahon et al., 2000), and it is well established that the re-ceptive field centers of midget ganglion cells in the retinal pe-riphery encompass an area sufficient to contain inputs from manycones (Dacey, 1993; Goodchild et al., 1996). Currently, there arealso continuing questions about the purity of the cone signal to thesurround regions of the red/green opponent cells and concern asto how these center/surround arrangements may be altered forganglion cells located in the retinal periphery. Whatever the eventualresolution of all these issues, it is clear that it is from signalsof cells of this type that a second dimension of color vision isextracted by the primate central visual system (Solomon & Lennie,2007).

The addition of a second class of M/L cone and the presence ofmidget cell circuitry permit the emergence of primate trichromacy.What is the evolutionary linkage between these two critical

628 Jacobs

elements? Since the presence of a second, spectrally discrete M/Lpigment will automatically expand the spectral window throughwhich an animal samples the photic environment, one possibilityis that pigment addition per se could yield visual advantages and,if so, the circuitry to compare outputs from the two pigment typesmight have been evolved subsequent to the addition of a newpigment type. A more popular idea, however, is that the pathwaysto channel signals from single cones to ganglion cells firstemerged as an adaptation to support higher spatial acuity. Withthis circuitry in place, the subsequent addition of a new M/L conetype would then automatically allow for an opponent comparisonof signals from M and L cones, thus supporting an immediate stepto trichromacy (e.g., Wassle & Boycott, 1991). Perhaps weighingagainst this possibility, Lee (2004) has pointed out that the spatialvision that can be derived from single-cone inputs to midgetganglion cells would exceed the resolution capacity of primateoptical systems. In his view, this could make it unlikely that themidget cell system, at least as it is currently configured in primateretinas, would have evolved solely as an adaptation to supporthigh spatial acuity.

Comparative studies shed some light on these issues, but theydo not resolve them. Comparisons of retinal ganglion cell anatomyand physiology in catarrhine and platyrrhine primates show thatthe retinal circuitry in representative species from the two groupsis very similar and that, further, except for differences in spectralsensitivity, dichromatic and trichromatic conspecific platyrrhinesalso have much the same midget cell organization (Kremers &Lee, 1998; Silveira et al., 2004). These observations imply that thepostreceptoral arrangements sufficient to support trichromaticcolor vision were already in place in the anthropoid ancestral tothese two lineages. Perhaps even more surprising, anatomicalstudies further suggest that a retinal midget cell system is alsopresent in the strepsirrhine bush baby (Yamada et al., 1998). Theapparent ubiquity of this system across contemporary primateswould imply that either this retinal organization arose long beforetrichromacy first emerged, perhaps to support some other visual needsuch as enhanced acuity, or, as has been suggested (Tan et al., 2005),trichromacy emerged earlier in primates than is usually believed.

Why among mammals are the primates uniquely trichromatic?

Either as a routine capacity or as a polymorphic alternative,trichromacy has been detected in animals from each of the threemajor primate groupings. As far as is known, however, no othereutherian mammals are trichromatic (Jacobs, 1993). What ac-counts for this disparity? It cannot be just a lack of sufficientcone population for, although some mammals indeed have sparsecone representation (e.g., comprising no more than ~1% of allphotoreceptors), a number of mammalian lineages feature retinaswith cone:rod ratios that exceed those of any primate—often verygreatly, as, for example, in the ground squirrel (Kryger et al.,1998) or the tree shrew (Muller & Peichl, 1989). Neither can it betraced entirely to photic lifestyle since, although there are to besure plenty of nocturnal mammals, there are numerous mammalsthat have been variously characterized as diurnal, crepuscular, orarrhythmic (Macdonald, 2001), nor does there seem any reason tobelieve that the likelihood of mutational changes in opsin genetuning sites that would lead to the emergence of new conepigments should have been significantly greater in the primatesthan in other mammals. These possibilities aside, the explanationmost frequently offered focuses on the absence of retinal midgetcell systems in nonprimate retinas. Boycott and Wassle (1999)

summarized the idea that a pigment gene mutation leading toseparate populations of L and M cones may not, by itself, besufficient to produce new color vision, by noting that ‘‘The splitmight result in equal numbers of L- and M-cones randomlydistributed across the retina. In mammals, other than primates,bipolar cells pool the signals of several neighboring cones and, inturn, ganglion cells pool the signals from several convergingbipolar cells. In such a highly convergent system the chromaticinformation introduced in the cone mosaic by the L/M mutationwould be lost within the retina and thus never reach the brain.’’ Ifthis idea is right, then even if new M/L pigments should appear innonprimate mammals, they might not evolve.

In recent years, attempts have been made to study this issuedirectly by using genetic engineering to introduce a second M/Lphotopigment into the retinas of mice, animals that, like mostother mammals, normally express only one such pigment (in thiscase, �max 5 510 nm). In a first attempt, mice were generated thatwere transgenic for a human L-cone opsin gene (linked to theproduction of a pigment with �max of 556 nm). In these animals,the transgene was incorporated autosomally with the result that~80% of the cones coexpressed human L- and the native mouseM-cone pigment (Shaaban et al., 1998). Spectral sensitivity curvesfor the full complement of cone pigments in these mice are shownin Fig. 7A. Electroretinogram (ERG) recordings showed that thehuman cone pigment worked efficiently in mouse cones, yieldingchanges in outer retinal signals consistent with the operational

Fig. 7. Results from experiments to evaluate the functional consequences

of adding a cone pigment to their normal pigment complement of the

mouse. (A) Spectral sensitivity curves for the three types of cone pigments

found in two types of bioengineered mice. The native pigments in the

mouse have �max values of 360 and 510 nm, while that for the added

pigment is 556 nm. (B) Behavioral increment threshold functions obtained

from wild-type mice (triangles) and transgenic mice (circles). In addition

to the two cone pigments found in wild-type mice, most cones of the latter

animal also expressed a human L-cone photopigment. The continuous lines

are best-fit linear summations of the absorption curves for cone pigments

found in each type of mouse. Results derived from Jacobs et al. (1999).

Primate color vision: A comparative perspective 629

presence of an L-pigment (Jacobs et al., 1999). Behavioral testsconducted to determine if the new pigment impacted the visualcapabilities of these mice showed that they do, as illustrated inFig. 7B, which shows increment threshold spectral sensitivityequivalently measured in wild-type and transgenic mice. Thesetransgenic mice acquired significantly enhanced sensitivity tolong-wavelength lights, much as would be predicted by the addedpresence of a cone pigment with a peak of 556 nm. A direct test toillustrate the magnitude of this change showed that these trans-genic mice could detect light at least 30 nm further into the longwavelengths than wild-type mice. Finally, although extensive testsfor color vision were run, the transgenic mice consistently failedto evidence any such capacity. Because the novel pigment wascoexpressed in cones containing the native pigments, the absenceof color vision is probably not surprising. Although no new colorvision emerged from the mere addition of a second M/L pigmentin this manner, clear alterations in visual sensitivity were pro-duced. As I have noted, changes like this might, under the rightcircumstances, provide an adaptive advantage.

To avoid the problem of pigment coexpression, two morerecent investigations targeted the human L-opsin gene to themouse X-chromosome in knock-in mice (Smallwood et al., 2003;Onishi et al., 2005). The gene/pigment arrangement in such micewas specifically designed to mimic that of the polymorphic NewWorld monkeys in that three separate M/L pigment phenotypes aregenerated by this procedure—mice having either the native M orhuman L pigment (all males and homozygous females) or animalsexpressing both M and L pigments in separate cone populations(heterozygous females). As in the case of the transgenic mice,ERG recordings showed that the human L-pigment efficientlytransduces light in mouse photoreceptors and that spectralsensitivity based on outer retinal signals was consistent withpredictions from the new opsin gene complements. Subsequently,extensive behavioral tests were conducted to establish if visionwas also altered in these mice (Jacobs et al., 2007). As with thetransgenic mice, knock-in mice expressing the human L conepigment showed predictable increases in long-wavelength sensi-tivity. Unlike the transgenic mice, however, the behavior ofseveral of the heterozygous knock-in mice indicated the presenceof a color vision capacity absent in control mice. Wavelengthdiscrimination functions derived for three such animals (Fig. 8)give one illustration of the nature of the new color vision.

The results illustrated in Fig. 8, along with additional colormatching tests, show quite clearly that color vision, as thatcapacity is formally defined, can be produced by the additionof a novel cone pigment to the normal cone complement ofa mammal. Three implications can be drawn from this demon-stration. First, these results reinforce the view that visual systemshave a remarkable plasticity in that a simple alteration in first-stage receptor molecules can, all by themselves, trigger theemergence of a complex new sensory capacity. Second, thepresence of a retinal midget cell system like that characteristicof primate retinas must not be an inescapable prerequisite foradding a new dimension of color vision. Third, as was suggestedearly in this line of research (Mollon et al., 1984), this resultimplies that the addition of a novel M/L cone pigment throughopsin gene alterations probably allowed our primate ancestors toimmediately gain new color vision.

These results also come with several caveats. First, despite thefact that all the heterozygous mice had populations of both M andL pigments in their retinas, only some animals seemed to acquirecolor vision (Jacobs et al., 2007). Whether this partial pattern of

success in the color vision tests has a mundane explanation (e.g.,some inadequacies in the training regimens) or whether it reflectssomething more basic is not known. Second, even those animalsthat did achieve novel color vision required thousands of trainingtrials before consistent color discriminations began to appear.What this may mean is also unclear; one possible interpretation isthat the color cues newly available to these knock-in mice are notvery salient, and thus, it is difficult to encourage the animal tomake use of them. Third, these experiments do not explicitlyaddress the question of how the mouse visual system is able toextract this new color signals. Many mouse ganglion cells havespatially antagonistic center/surround receptive fields (Sagdullaev& McCall, 2005), so one possibility is that in these knock-in micethere were sufficient differences in the L and M cone represen-tation into these two regions to yield spectrally opponent signals.Alternatively, some upstream comparison of the relative M and Lsignal weightings across ganglion cell populations might alsoallow for the extraction of a color signal. Well-directed recordingexperiments could reveal just how these heterozygous mice extractand exploit the new color signal.

Given that these knock-in mice were able to add new colorvision by exploiting the artificially induced presence of a new typeof cone pigment, one wonders why a similar change did nothappen in some mammalian lineages during the course of theirevolution. To start with, we cannot be absolutely confident that ithas not occurred, being present somewhere among that significantcohort of mammals whose cone photopigment and color visionstatus remain to this day uninvestigated. And there is also thepossibility that even if an opsin gene change had yielded a newcone pigment and the nervous system had been capable ofgenerating new color vision, that capacity may not have alteredspecies fitness sufficiently for it to be maintained over subsequentgenerations. Another possibility is that in the absence of a midgetcell system, cone and ganglion cell density may critically in-fluence the prospects for extracting a spectrally opponent signal.

Fig. 8. Wavelength discrimination functions for heterozygous knock-in

mice whose retinas contain all three of the photopigments whose spectra

are plotted in Fig. 7A. Each point plots the difference in wavelength (in

nanometers) required for successful discrimination at each of the seven

spectral locations. These results illustrate the clear presence of novel color

vision in these mice that is traceable to the presence of the added cone

photopigment. The separate symbols represent results from different

animals (data from Jacobs et al., 2007).

630 Jacobs

For instance, mammals having very high cone densities may beunlikely candidates for acquiring new color vision in this waysince, much in the fashion described by Boycott and Wassle(quoted above), the mixed convergence of signals from the twopigment types would minimize the possibility of allowing strongganglion cell spectral opponency. Somewhat paradoxically, asparser cone representation may provide a more favorable sub-strate for generating a novel color signal in these retinas. In theknock-in mouse, for example, X-chromosome inactivation re-sulted in the appearance of patch-like clusters of M and L conescontaining the same pigment type (Smallwood et al., 2003; Onishiet al., 2005). The patch sizes so generated are not much smallerthan the graininess of ganglion cell receptive fields, and thisapproximate matching may have been critical for permittinguseful spectral comparisons between the two cone types.

Although we do not yet know in detail what factors mayeventually be required to explain why trichromatic color visionemerged in primates but not in other mammals, it seems indisput-able that the primate midget cell provides a remarkably efficientneural substrate for setting up the initial comparisons betweendifferent cone types that are required to support M/L color vision.Whether these pathways initially evolved as an adaptation tosupport higher spatial acuity or whether they evolved in conjunc-tion with very early primate trichromacy, the primate midget cellsystem has fostered flexibility for color vision change in primatesthat is not available in other mammalian visual systems.

Acknowledgments

Much of the material presented in this paper was developed for two recenttalks—the Robert M. Boynton Lecture given at the Fall Vision Meeting ofthe Optical Society of America (2007) and the Russell De Valois MemorialLecture presented at the School of Optometry, University of California,Berkeley (2008). I thank the organizers of the lectures for inviting me toparticipate in these recurring events that were initiated to honor thecontributions made by these two outstanding vision scientists, both ofwhom were instrumental in advancing our understanding of primate colorvision. Useful comments made by two reviewers and support from theNational Eye Institute (EY002052) are gratefully acknowledged. Kris Kroghhelped in the preparation of figures and composed the cover illustration.

References

Arrese, C.A., Beazley, L.D. & Neumeyer, C. (2006). Behaviouralevidence of marsupial trichromacy. Current Biology 16, R193–R194.

Arrese, C.A., Hart, N.S., Thomas, N., Beazley, L.D. & Shand, J.(2002). Trichromacy in Australian marsupials. Current Biology 12,657–660.

Bearder, S.K., Nekaris, K.A.I. & Curtis, D.J. (2006). A re-evaluationof the role of vision in the activity and communication of nocturnalprimates. Folia Primatologica 77, 50–71.

Bininda-Emonds, O.R.P., Cardillo, M., Jones, K.E., MacPhee, R.D.E.,Beck, R.M.D., Grenyer, R., Price, S.A., Vos, R.A., Gittleman, J.L.& Purvis, A. (2007). The delayed rise of present-day mammals.Nature 446, 507–512.

Boissinot, S., Tan, Y., Shyue, S.-K., Schneider, H., Sampaio, I.,Neiswanger, K., Hewett-Emmett, D. & Li, W.-H. (1998). Originsand antiquity of X-linked triallelic color vision systems in New Worldmonkeys. Proceedings of the National Academy of Sciences U S A 95,13749–13754.

Bowmaker, J.K. (1990). Cone visual pigments in monkeys and humans.In Advances in Photoreception, ed. Committee on Vision, pp. 19–30.Washington, DC: National Academy Press.

Bowmaker, J.K. (2008). Evolution of vertebrate visual pigments. VisionResearch 48, 2022–2041.

Bowmaker, J.K., Astell, S., Hurst, D.M. & Mollon, J.D. (1991).Photosensitive and photostable pigments in the retinae of Old Worldmonkeys. Journal of Experimental Biology 156, 1–19.

Boycott, B. & Wassle, H. (1999). Parallel processing in the mammalianretina. Investigative Ophthalmology and Visual Science 40, 1313–1327.

Brown, P.K. & Wald, G. (1963). Visual pigments in human and monkeyretinas. Nature 200, 37–43.

Buck, S.L. (2003). Rod-cone interactions in human vision. In The VisualNeurosciences, Vol. 1, ed. Chalupa, L.M. & Werner, J.S., pp. 863–878. Cambridge, MA: MIT Press.

Buck, S.L., Knight, R., Fowler, G. & Hunt, B. (1998). Rod influenceon hue-scaling functions. Vision Research 38, 3259–3263.

Calkins, D.J. (2001). Seeing with S cones. Progress in Retinal and EyeResearch 20, 255–287.

Cao, D., Poicorny, J. & Smith, V.C. (2005). Matching rod percepts withcone stimuli. Vision Research 45, 2119–2128.

Collin, S.P. & Trezise, A.E.O. (2004). The origins of colour vision invertebrates. Clinical and Experimental Optometry 87, 217–233.

Dacey, D.M. (1993). The mosaic of retinal ganglion cells in the humanretina. Journal of Neuroscience 13, 5324–5355.

Dacey, D.M. & Lee, B.B. (1994). The ‘‘blue-on’’ opponent pathway inprimate retina originates from a distinct bistratified ganglion cell type.Nature 367, 731–735.

Darwin, C. (1859). On the Origin of Species. London: John Murray.Davies, W.L., Carvalho, L.S., Cowing, J.A., Beazley, L.D.,

Hunt, D.M. & Arrese, C. (2007). Visual pigments of the platypus:A novel route to mammalian colour vision. Current Biology 17,R161–R163.

Dominy, N.J. (2004). Color as an indicator of food quality to anthropoidprimates: Ecological evidence and an evolutionary scenario. InAnthropoid Origins: New Visions, ed. Ross, C.F. & Kay, R.F., pp.615–644. New York: Kluwer Academic/Plenum Publishers.

Dulai, K.S., von Dornum, M., Mollon, J.D. & Hunt, D.M. (1999). Theevolution of trichromatic color vision by opsin gene duplication inNew World and Old World primates. Genome Research 9, 629–638.

Fernandez-Duque, E. (2003). Influences of moonlight, ambient temper-ature, and food availability on the diurnal and nocturnal activity of owlmonkeys (Aotus azarai). Behavioral Ecology and Sociobiology 54,431–440.

Fletcher, R. & Voke, J. (1985). Defective Colour Vision: Fundamentals,Diagnosis and Management. Bristol, CT: Adam Hilger Ltd.

Gimenez, M. & Fernandez-Duque, E. (2003). Summer and winter dietof night monkeys in the gallery and thorn forests of the ArgentineanChaco. Revista de Etologia 5,Supplement 164.

Goodchild, A.K., Gosh, K.K. & Martin, P.R. (1996). Comparison ofphotoreceptor spatial density and ganglion cell morphology in theretina of human, macaque monkey, cat, and the marmoset (Callithrixjacchus). Journal of Comparative Neurology 366, 55–75.

Hart, N.S. & Hunt, D.M. (2007). Avian visual pigments: Characteristics,spectral tuning, and evolution. The American Naturalist 169, Supple-ment, S70/a26.

Haverkamp, S., Wassle, H., Duebel, J., Kuner, T., Augustine, G.J.,Feng, G. & Euler, T. (2005). The primordial, blue-cone color systemof the mouse retina. Journal of Neuroscience 25, 5438–5445.

Heesy, C.P. & Ross, C.F. (2001). Evolution of activity patterns andchromatic vision in primates: Morphometrics, genetics and cladistics.Journal of Human Evolution 40, 111–149.

Hendrickson, A., Djajadi, H.R., Nakamura, L., Possin, D.E. &Sajuthi, D. (2000). Nocturnal tarsier retina has both short and long/medium-wavelength cones in an unusual topography. Journal ofComparative Neurology 424, 718–730.

Hunt, D.M. (2006). Molecular evolution of colour vision in primates.Journal of Vision 6, 34a.

Hunt, D.M., Cowing, J.A., Wilkie, S.E., Parry, J.W.L., Poopala-

sundaram, S. & Bowmaker, J.K. (2004). Divergent mechanisms forthe tuning of shortwave sensitive visual pigments in vertebrates.Photochemical and Photobiological Sciences 3, 713–720.

Hunt, D.M., Jacobs, G.H. & Bowmaker, J.K. (2005). The genetics andevolution of primate visual pigments. In The Primate Visual System:A Comparative Approach, ed. Kremers, J., pp. 73–126. Chichester,UK: John Wiley & Sons Ltd.

Hunt, D.M., Williams, A.J., Bowmaker, J.K. & Mollon, J.D. (1993).Structure and evolution of polymorphic photopigment gene of themarmoset. Vision Research 33, 147–154.

Jacobs, G.H. (1993). The distribution and nature of colour vision amongthe mammals. Biological Reviews 68, 413–471.

Jacobs, G.H. (1996). Primate photopigments and primate color vision.Proceedings of the National Academy of Sciences U S A 93, 577–581.

Primate color vision: A comparative perspective 631

Jacobs, G.H. (1998). A perspective on color vision in platyrrhine monkeys.Vision Research 38, 3307–3313.

Jacobs, G.H. (2007). New World monkeys and color. InternationalJournal of Primatology 28, 729–759.

Jacobs, G.H. & Deegan, J.F., II. (1993). Photopigments underlying colorvision in ringtail lemurs (Lemur catta) and brown lemurs (Eulemurfulvus). American Journal of Primatology 30, 243–256.

Jacobs, G.H. & Deegan, J.F., II. (1999). Uniformity of colour vision inOld World monkeys. Proceedings of the Royal Society of London B266, 2023–2028.

Jacobs, G.H. & Deegan, J.F., II. (2005). Polymorphic monkeys with morethan three M/L cone types. Journal of the Optical Society of America A22, 2072–2080.

Jacobs, G.H., Deegan, J.F., II, Neitz, J.A., Crognale, M.A. & Neitz,M. (1993). Photopigments and color vision in the nocturnal monkey,Aotus. Vision Research 33, 1773–1783.

Jacobs, G.H., Deegan, J.F., II, Tan, Y. & Li, W.-H. (2002). Opsin geneand photopigment polymorphism in a prosimian primate. VisionResearch 42, 11–18.

Jacobs, G.H., Fenwick, J.C., Calderone, J.B. & Deeb, S.S. (1999).Human cone pigment expressed in transgenic mice yields alteredvision. Journal of Neuroscience 19, 3258–3265.

Jacobs, G.H., Neitz, M., Deegan, J.F. & Neitz, J. (1996a). Trichromaticcolour vision in New World monkeys. Nature 382, 156–158.

Jacobs, G.H., Neitz, M. & Neitz, J. (1996b). Mutations in S-conepigment genes and the absence of colour vision in two species of nocturnalprimate. Proceedings of the Royal Society of London B 263, 705–710.

Jacobs, G.H. & Williams, G.A. (2001). The prevalence of defective colorvision in Old World monkeys and apes. Color Research and Applica-tion 26, S123–S127.

Jacobs, G.H. & Williams, G.A. (2006). L and M cone proportions inpolymorphic New World monkeys. Visual Neuroscience 23, 365–370.

Jacobs, G.H., Williams, G.A., Cahill, H. & Nathans, J. (2007).Emergence of novel color vision in mice engineered to express a humancone photopigment. Science 315, 1723–1725.

Kainz, P.M., Neitz, J. & Neitz, M. (1998). Recent evolution of uniformtrichromacy in a New World monkey. Vision Research 38, 3315–3320.

Kawamura, S. & Kubotera, N. (2004). Ancestral loss of short wave-sensitive cone visual pigment in lorsiform prosimians, contrasting withits strict conservation in other prosimians. Journal of MolecularEvolution 58, 314–321.

Kelber, A. & Roth, L.S.V. (2006). Nocturnal colour vision—Not as rareas we might think. Journal of Experimental Biology 209, 781–788.

Kielan-Jaworowska, Z., Cifelli, R.L. & Luo, Z.-X. (2004). Mammalsfrom the Age of Dinosaurs: Origins, Evolution, and Structure. NewYork: Columbia University Press.

Kremers, J. & Lee, B.B. (1998). Comparative retinal physiology inAnthropoids. Vision Research 38, 3339–3344.

Kryger, Z., Galli-Resta, L., Jacobs, G.H. & Reese, B.E. (1998). Thetopography of rod and cone photoreceptors in the retina of the groundsquirrel. Visual Neuroscience 15, 685–691.

Lee, B.B. (2004). Paths to colour in the retina. Clinical and ExperimentalOptometry 87, 239–248.

Levenson, D.H., Fernandez-Duque, E., Evans, S. & Jacobs, G.H.(2007). Mutational changes in S-cone opsin genes common to bothnocturnal and cathemeral Aotus monkeys. American Journal ofPrimatology 69, 757–765.

Li, W. & DeVries, S.H. (2006). Bipolar cell pathways for color and humanvision in a dichromatic mammalian retina. Nature Neuroscience 9, 669–675.

Macdonald, D., ed. (2001). The New Encyclopedia of Mammals. Oxford:Oxford University Press.

MacNeil, M.A. & Gaul, P.A. (2008). Biocytin wide-field bipolar cells inrabbit retina selectively contact blue cones. Journal of ComparativeNeurology 506, 6–15.

Makous, W. (2004). Scotopic vision. In The Visual Neurosciences, Vol. 1, ed.Chalupa, L.M. & Werner, J.S., pp. 838–850. Cambridge, MA: MIT Press.

Marks, W.B., Dobelle, W.H. & MacNichol, E.F.J. (1964). Visualpigments of single primate cones. Science 143, 1181–1183.

Martin, P.R. (1998). Colour processing in the retina: Recent progress.Journal of Physiology 513, 631–638.

Martin, R.D. (1990). Primate Origins and Evolution. Princeton, NJ:Princeton University Press.

Martin, R.D. & Ross, C.F. (2005). The evolutionary and ecologicalcontext of primate vision. In The Primate Visual System: A ComparativeApproach, ed. Kremers, J. West Sussex, UK: John Wiley & Sons, Ltd.

Martin, R.D., Soligo, C. & Tavare, S. (2007). Primate origins:Implications of a cretaceous ancestry. Folia Primatologica 78, 277–296.

Masland, R.H. (2001). The fundamental plan of the retina. NatureNeuroscience 4, 877–886.

McMahon, M.J., Lankheet, M.J., Lennie, P. & Williams, D.A. (2000).Fine structure of parvocellular receptive fields in the primate fovea revealedby laser interferometry. Journal of Neuroscience 20, 2043–2053.

Mollon, J.D. (2003). The origins of modern color science. In The Scienceof Color, ed. Shevell, S.K., pp. 1–39. Amsterdam, The Netherlands:Elsevier.

Mollon, J.D., Bowmaker, J.K. & Jacobs, G.H. (1984). Variations ofcolour vision in a New World primate can be explained by poly-morphism of retinal photopigments. Proceedings of the Royal Societyof London B 222, 373–399.

Muller, B. & Peichl, L. (1989). Topography of cones and rods in the treeshrew retina. Journal of Comparative Neurology 282, 581–594.

Nathans, J., Thomas, D. & Hogness, D.S. (1986). Molecular genetics ofhuman color vision: The genes encoding blue, green and red pigments.Science 232, 193–202.

Neitz, M. & Neitz, J. (2003). Molecular genetics of human color visionand color vision defects. In The Visual Neurosciences, Vol. 2, ed. Chalupa,L.M. & Werner, J.S., pp. 974–988. Cambridge, MA: MIT Press.

Neitz, M., Neitz, J. & Jacobs, G.H. (1991). Spectral tuning of pigmentsunderlying red-green color vision. Science 252, 971–974.

Ni, X., Wang, Y., Hu, Y. & Li, C. (2003). A euprimate skull from the earlyEocene of China. Nature 427, 65–68.

Onishi, A., Hasegawa, J., Imai, H., Chisaka, O., Ueda, Y., Honda, Y.,Tachibana, M. & Shichida, Y. (2005). Generation of knock-in micecarrying third cones with spectral sensitivity different from S and Lcones. Zoological Science 22, 1145–1156.

Onishi, A., Koike, S., Ida, M., Imai, H., Schichida, Y., Osamu, T.,Hanazawa, A. Konatsu, H., Mikami, A., Goto, S., Suryobroto, B.,Kitahara, K. & Yamamori, T. (1999). Dichromatism in macaquemonkeys. Nature 402, 139–140.

Osorio, D., Smith, A.C., Vorobyev, M. & Buchanan-Smith, H.M.(2004). Detection of fruit and the selection of primate visual pigmentsfor color vision. American Naturalist 164, 696–708.

Peichl, L. (2005). Diversity of mammalian photoreceptor properties: Adap-tations to habitat and lifestyle? Anatomical Record A 287A, 1001–1012.

Perry, G.H., Martin, R.D. & Verelli, B.C. (2007). Signatures offunctional constraint at aye-aye opsin genes: The potential of adaptivecolor vision in a nocturnal primate. Molecular Biology and Evolution24, 1963–1970.

Post, R.H. (1962). Population differences in red and green color visiondeficiency: A review, and query on selection relaxation. EugenicsQuarterly 9, 131–146.

Regan, B.C., Julliot, C., Simmen, B., Vienot, F., Charles-Dominique,P. & Mollon, J.D. (2001). Fruits, foliage and the evolution of primatecolour vision. Philosophical Transactions of the Royal Society ofLondon B 356, 229–283.

Ross, C.F. & Martin, R.D. (2007). The role of vision in the origin andevolution of primates. In Evolution of Nervous Systems. Vol. 4: TheEvolution of Primate Nervous Systems, ed. Preuss, T.M. & Kaas, J.,pp. 59–78. Oxford: Elsevier.

Sagdullaev, B.T. & McCall, M.A. (2005). Stimulus size and intensityalter fundamental receptive field properties of mouse retinal ganglioncells in vivo. Visual Neuroscience 22, 649–659.

Saito, C.A., da Silva-Filho, M., Lee, B.B., Bowmaker, J.K., Kremers,J. & Silveira, L.C.L. (2004). Alouatta trichromatic color vision—Single-unit recording from retinal ganglion cells and micro spectro-photometry. Investigative Ophthalmology and Visual Science 45,E-abstract 4276.

Schmitz, J., Ohme, M. & Zischler, H. (2001). SINE insertions incladistic analyses and the phylogenetic affiliations of Tarsius bancanusto other primates. Genetics 157, 777–784.

Schrago, C.G. (2007). On the time scale of New World primate di-versification. American Journal of Physical Anthropology 132, 344–354.

Seiffert, E.R., Simons, E.L., Clyde, W.C., Rossie, J.B., Attia, Y.,Brown, T.M., Chatrath, P. & Mathison, M.E. (2005). Basalanthropoids from Egypt and the antiquity of Africa’s higher primateradiation. Science 310, 300–304.

Shaaban, S.A., Crognale, M.A., Calderone, J.B., Huang, J., Jacobs,G.H. & Deeb, S.S. (1998). Transgenic mice expressing a functionalhuman photopigment. Investigative Ophthalmology and Visual Science39, 1036–1043.

632 Jacobs

Silveira, L.C.L., Lee, B.B., Yamada, E.S., Kremers, J., Hunt, D.M.,Martin, P.R. & Gomes, F.L. (1999). Ganglion cells of short-wave-length-sensitive pathway in New World monkeys: Morphology andphysiology. Visual Neuroscience 16, 333–343.

Silveira, L.C.L., Saito, C.A., Lee, B.B., Kremers, J., Filho, M.S.,Kilavik, B.E., Yamada, E.S. & Perry, V.H. (2004). Morphology andphysiology of primate M- and P-cells. In The Roots of VisualAwareness, ed. Heywood, C.A., Milner, A.D. & Blakemore, C.,pp. 21–46. Amsterdam, The Netherlands: Elsevier.

Smallwood, P.M., Olveczky, B.P., Williams, G.A., Jacobs, G.H.,Reese, B.E., Meister, M. & Nathans, J. (2003). Geneticallyengineered mice with an additional class of cone photoreceptors:Implications for the evolution of color vision. Proceedings of theNational Academy of Sciences U S A 100, 11706–11711.

Smallwood, P.M., Wang, Y. & Nathans, J. (2002). Role of a locuscontrol region in the mutually exclusive expression of human red andgreen cone pigment genes. Proceedings of the National Academy ofSciences U S A 99, 1008–1011.

Smith, V.C. & Pokorny, J. (1977). Large-field trichromacy in protanopesand deuteranopes. Journal of the Optical Society of America 67, 213–220.

Smith, V.C. & Pokorny, J. (2003). Color matching and color discrimi-nation. In The Science of Color, ed. Shevell, S.K., pp. 103–148.Amsterdam, The Netherlands: Elsevier.

Solomon, S.G. & Lennie, P. (2007). The machinery of colour vision.Nature Neuroscience Reviews 8, 276–286.

Steiper, M.E. & Young, N.M. (2006). Primate molecular divergencedates. Molecular Phylogenetics and Evolution 41, 384–394.

Surridge, A.K. & Mundy, N.I. (2002). Trans-specific evolution of opsinalleles and the maintenance of trichromatic colour vision in callitri-chine primates. Molecular Ecology 11, 2157–2169.

Tan, Y. & Li, W.-H. (1999). Trichromatic vision in prosimians. Nature402, 36.

Tan, Y., Yoder, A.D., Yamashita, N. & Li, W.-H. (2005). Evidencefrom opsin genes rejects nocturnality in ancestral primates.Proceedings of the National Academy of Sciences U S A 41, 14712–14716.

Tattersal, I. (2006). The concept of cathemerality: History and definition.Folia Primatologica 77, 7–14.

Vorobyev, M. (2004). Ecology and evolution of primate colour vision.Clinical and Experimental Optometry 87, 230–238.

Walls, G.L. (1942). The Vertebrate Eye and Its Adaptive Radiation.Bloomfield Hills, MI: Cranbrook Institute of Science.

Wang, Y., Macke, J.P., Merbs, S.L., Zack, D.J., Klaunberg, B.,Bennett, J., Gearhart, J. & Nathans, J. (1992). A locus controlregion adjacent to the human red and green visual pigment genes.Neuron 9, 429–440.

Wassle, H. (2004). Parallel processing in the mammalian retina. NatureNeuroscience 19, 747–757.

Wassle, H. & Boycott, B.B. (1991). Functional architecture of themammalian retina. Physiological Reviews 71, 447–480.

Wyszecki, G. & Stiles, W.S. (1982). Color Science. New York: Wiley.Yamada, E.S., Marshak, D.W., Silveira, L.C.L. & Casagrande, V.A.

(1998). Morphology of P and M retinal ganglion cells of the bush baby.Vision Research 38, 3345–3352.

Yeh, T., Lee, B.B., Kremers, J., Cowing, J.A., Hunt, D.M., Martin,P.R. & Troy, J.B. (1995). Visual responses in the lateral geniculatenucleus of dichromatic and trichromatic marmosets (Callithrixjacchus). Journal of Neuroscience 15, 7892–7904.

Yokoyama, S. (2000). Molecular evolution of vertebrate visual pigments.Progress in Retinal and Eye Research 19, 385–419.

Yokoyama, S. & Radlwimmer, F.B. (1999). The molecular genetics ofred and green color vision in mammals. Genetics 153, 919–932.

Zhang, J. (2003). Evolution by gene duplication: An update. Trends inEcology and Evolution 18, 292–298.a

Primate color vision: A comparative perspective 633